Abstract
99mTc-methoxyisobutylisonitrile (99mTc-MIBI) has been suggested as a tracer for the scintigraphic detection of multidrug resistance (MDR). The aim of this study was to compare MDR characteristics in vitro and in vivo by immunohistochemic and functional uptake assays in established tumor cell lines cultured and grown in severe combined immunodeficient (SCID) mice. Methods: The presence of MDR was assessed in vitro in drug-resistant HT-29mdr1 colon carcinoma cells and in nonresistant HT-29par cells by JSB-1 immunohistochemistry, uptake of the fluorescent dye Rhodamine 123, and quantitative measurement of 99mTc-MIBI accumulation. For in vivo imaging, SCID mice bearing subcutaneous xenografts of these cell lines were injected with 99mTc-MIBI and 18F-FDG for scintigraphic and PET examination. After imaging, tumors were analyzed by immunohistochemistry and electron microscopy. Results: All HT-29mdr1 cells cultured in vitro exhibited distinct JSB-1 immunoreactivity, although to a variable degree, whereas HT-29par cells were completely devoid of JSB-1 staining. Rhodamine 123 accumulated poorly in HT-29mdr1 cells but strongly in HT-29par cells. Accumulation of 99mTc-MIBI was 0.05% ± 0.01% of the activity of the external medium in HT-29mdr1 cells, but about eight times higher in HT-29par cells (0.40% ± 0.09%), a very low percentage compared with other tumor cell lines. No difference in 201TlCl accumulation was observed between both cell lines. In vivo, neither HT-29par nor HT-29mdr1 tumors grown in SCID mice could be detected by 99mTc-MIBI scintigraphy. In FDG PET, both HT-29mdr1 and HT-29par tumors were clearly visible. FDG uptake was, however, markedly higher in HT-29par than in HT-29mdr1 tumors. Both tumor types were poorly vascularized, as shown histologically. JSB-1 immunoreactivity was absent in all HT-29par tumors examined, whereas the majority of HT-29par tumor cells were stained. Electron microscopy showed that HT-29par tumors contained significantly less mitochondria than hepatocytes of the SCID mouse liver, which displayed high 99mTc-MIBI uptake in our scintigraphy studies. Conclusion: Sufficient 99mTc-MIBI uptake is the major prerequisite for distinguishing successfully between drug-resistant and sensitive cells. Negative 99mTc-MIBI scintigrams are not necessarily associated with MDR expression. In some tumors, FDG may be an in vivo marker for MDR as suggested by PET.
- human HT-29 colon cancer cell line
- electron microscopy
- FDG
- PET
- immunohistochemistry
- multidrug resistance
- Rhodamine 123
- severe combined immunodeficient mouse
- 99mTc-methoxyisobutylisonitrile
- 201TlCl
One major obstacle in the chemotherapeutic treatment of cancer patients is the emergence of tumor cells resistant to anticancer agents. The lack of response often develops in tumors, which initially have responded well to chemotherapy, simultaneously involving various different chemotherapeutic agents, such as anthracyclins (e.g., doxorubicine), alkaloids (e.g., vincristine, colchicine), and epipodophyllotoxins (VP-16). This acquired resistance to a wide range of unrelated drugs is referred to as multidrug resistance (MDR). The mechanism most frequently involved is the overexpression of a 170-kDa plasma membrane phosphoglycoprotein, P-gp, encoded by the MDR1 gene (1,2). P-gp acts as a transmembrane energy-dependent drug-efflux pump that transports several apparently unrelated organic compounds, such as cytostatics, out of the cell, resulting in drug resistance (3).
To design the most efficient therapy protocols and to reduce unwanted secondary effects of chemotherapy to a minimum, it is of great clinical importance to predict the outcome of cancer therapy by identifying those patients that will not respond to anticancer treatment. 99mTc-methoxyisobutylisonitrile (99mTc-MIBI) is a suitable transport substrate of P-gp, and its cellular accumulation is inversely proportional to the level of P-gp expression (4,5). It has therefore been hypothesized that functional in vivo imaging with 99mTc-MIBI may allow the rapid characterization of P-gp expression in tumors (5). This procedure would permit us to assess the efficacy of chemotherapy and thus to select appropriate chemotherapy regimens. In fact, several reports suggest that tumors with higher 99mTc-MIBI uptake are more likely to respond to chemotherapy than those with a lower uptake (6,7). Other studies, however, indicate that 99mTc-MIBI imaging is not useful for predicting the response to chemotherapy (8). Further investigations suggest that P-gp expression is not exclusively related to initial 99mTc-MIBI uptake, but rather to the washout of 99mTc-MIBI from the tumor (9).
These contradictory results may, in part, be related to the fact that different forms of human tumors have been studied. We have therefore chosen an established human colon carcinoma cell line (HT-29) and its MDR counterpart (10–12) to study P-gp expression by the same cancer cell line in vitro and in vivo. The severe combined immunodeficient (SCID) mouse model allows the systematic analysis of in vivo characteristics of human tumors growing in the microenvironment of a living organism (13,14). Human cancer cells cultured in vitro are transplanted into SCID mice, which lack functional B and T lymphocytes. In this system, tumor cells retain the morphology and functional characteristics of the original tumor and thereby rather accurately reflect the clinical situation (14).
Our purpose was to determine P-gp expression of nonresistant and MDR HT-29 human colon carcinoma cell lines both in vitro by immunohistochemic and functional assays and in vivo by functional imaging using 99mTc-MIBI and 18F-FDG in the same carcinoma cells growing in SCID mice. We then compared these results with histologic and electron microscopic analyses.
MATERIALS AND METHODS
Cell Culture
The human colon carcinoma cell line HT-29par was obtained from the American Type Culture Collection through the European Tissue Culture Collection (Porton Down, Salisbury, U.K.). The MDR HT-29mdr1 cell line (10) was generously supplied by Dr. I. N. Slotki (University of Jerusalem, Israel). Cells were cultivated to confluence in 50-mL tissue culture flasks under standard conditions (37°C, 100% relative humidity, 5% CO2/95% air) in McCoy’s 5A medium (Gibco/Life Technologies, Karlsruhe, Germany), supplemented with 10% heat-inactivated fetal calf serum ([FCS] Gibco), 2-mM l-glutamine, 100 U/mL penicillin, and 100 μg/mL streptomycin (cell culture medium). In addition, for maintaining the HT-29mdr1 cell line, 300 ng/mL colchicine was added to the culture medium. Because substantial numbers of cells were necessary to standardize and reproduce the experiments, cells were subcultured and cryopreserved in a cryo-safe I (c.c. pro GmbH, Neustadt, Germany) medium in multiple identical aliquots. Before each experiment, cells were thawed and incubated for 7 to 14 d in the culture medium.
For immunohistochemic detection of MDR, HT-29par and HT-29mdr1 cells cultivated on Chamber Slides (Nunc, Naperville, IL) were rinsed in 0.1 mol/L phosphate-buffered saline ([PBS] 4°C, pH 7.4) and then fixed in 4% paraformaldehyde in 0.1 mol/L phosphate buffer (1 h, 4°C, pH 7.4). After several rinses in PBS, cells were processed for immunohistochemistry as described below.
In addition, MDR was visualized by an assay based on an increased rate of fluorescent dye efflux in MDR cells, resulting in reduced staining (15). After cultivation for 48 h on Chamber Slides, HT-29par and HT-29mdr1 cells were incubated for 10 min with 5 μg/mL Rhodamine 123 (Sigma, Munich, Germany) in the culture medium (37°C, 5% CO2). To exclude different mechanisms of reduced Rhodamine uptake, the MDR transporter was inhibited by adding 10 μg/mL weak detergent Tween 80 (Fluka, Buchs, Switzerland) to the medium in control incubations. After several rinses in the cell culture medium, coverslips were placed over the slides using Crystal Mount plus Clarion Permanent Mounting Media (Biomeda Corp., Foster City, CA), examined immediately, and photographed under a Zeiss Axiophot fluorescence microscope (Carl Zeiss, Oberkochem, Germany).
Uptake of 99mTc-MIBI In Vitro
For quantitative assessment of the MDR transporter function, 99mTc-MIBI was applied to cultured tumor cells, and its uptake was compared with that of 201TlCl (DuPont Pharma, Brussels, Belgium). Synthesis of 99mTc-MIBI was performed with a one-step kit formulation (Cardiolite; DuPont, Bad Homburg v.d.H., Germany) as described by Piwnica et al. (16). Radiochemical purity was ≥95%, as measured by thin-layer chromatography using aluminum oxide plates (1B-F; JT Baker, Phillipsburg, NJ).
Adherent cells were harvested with 0.05% trypsin/0.53-mM EDTA (Gibco) and washed once in the culture medium. After a recovery period (incubation for 2 h at 37°C; 5% CO2), cells were counted on a Neubauer hemocytometer (Optik Labor, Friedrichshofen, Germany), suspended in the same serum-free medium at a concentration of 0.3 × 106 cells/mL, and distributed into 18 test tubes each (for HT-29par and HT-29mdr1 cells), 3 mL per tube. Two test tubes containing the culture medium, but no cells, served as control tubes.
A volume of 50 μL 99mTc-MIBI, dissolved in 0.9% NaCl to a final concentration of 20 kBq/50 μL, was added to nine test tubes and one control tube. Specific activity of 99mTc taken from the generator was about 1 MBq/nmol; that is, the total amount of MIBI was about 20 pmol/106 cells, which is far below saturation (17). The remaining nine test tubes and the remaining control tube were incubated with 50 μL 201TlCl and dissolved in 0.9% NaCl to a final concentration of 20 kBq/50 μL. The specific activity of 201Tl was about 0.7 MBq/nmol; that is, the total amount of 201TlCl was about 30 pmol/106cells. After an incubation period of 1 h (37°C, 5% CO2), tracer uptake by the cells was stopped by adding 7 mL ice-cold water, thus limiting the duration of exposition to the tracers to a defined period. After three rinses in ice-cold PBS (5 min at 1,500 rpm), cell pellets were suspended in 1 mL ice-cold PBS in Eppendorf tubes (Eppendorf, Hamburg, Germany). The radioactivity of the pellets was determined in a gamma well counter (FH 412; Frieseke & Hoepfner, Erlangen-Bruck, Germany). The Eppendorf tubes were measured for 1 min within the 99mTc energy window and for 1 min within the 201Tl energy window (window width, 40–100 mV; high voltage at maximum counting rate). Measured counting rates were corrected for background and radioactive decay. To obtain relative tracer uptake, the counting rate of the cell probes was divided by the counting rate of the appropriate control probe. The two-tailed Student t test for unpaired data was used to evaluate statistical differences between the tracer uptake of HT-29par and HT-29mdr1 cells. The t test was performed for equal (homoscedastic) or unequal (heteroscedastic) variances according to the result of Levene’s test.
To minimize nonspecific binding of 99mTc-MIBI to plastic or metal surfaces, glassware was used whenever possible. Plastic pipette tips were presaturated with freshly prepared 1% bovine serum albumin (A9645; Sigma, Deisenhofen, Germany) in 0.1 mol/L PBS for 1 h (pH 7.4) followed by three washes in PBS (17).
Animals
Twenty-one male adult (10–14-wk-old) pathogen-free BALB/c SCID mice were used in this study. Principles of laboratory animal care (NIH publication no. 86–23, revised 1985) were followed, as well as the German Law on Animal Protection from 1987, and the local health ethics committee approved the experiments. Animals were kept in filter-top cages and were provided with sterile water and standard food (ssniff M-Z Alleindiät extrudiert; ssniff Spezialdiäten GmbH, Soest, Germany) ad libitum. All manipulations were performed aseptically under a laminar flow hood. For injection, HT-29par and HT-29mdr1 cells were harvested by trypsinization and tested for viability (95%) after a culture period of 7–14 d, and 5 × 106 viable cells were resuspended in 1 mL McCoy’s 5A medium (Gibco). Each recipient SCID mouse was injected subcutaneously with 200 μL of the cell suspension into the back between the scapulae; 12 animals received HT-29par and 9 animals received HT-29mdr1 cells. When solid tumors had grown to a size of 1–1.5 cm3 (after 3 wk for HT-29par and after 5 wk for HT-29mdr1), the animals underwent various imaging procedures.
Uptake of 99mTc-MIBI In Vivo
In each SCID mouse, about 200 μL 99mTc-MIBI, dissolved in 0.9% NaCl to a final concentration of 100 MBq/200 μL, were administered in the tail vein using a 1-mL Luer syringe (Braun, Melsungen, Germany) with a Microlance canula (Becton Dickinson, Dublin, Ireland). Planar scintigraphic images with a pixel size of 1.0 × 1.0 mm2 were acquired over 15 min at 15 and 60 min after injection of 99mTc-MIBI. Each image was taken from the posterior view using a single-head gamma camera (Diacam; Siemens Medical Systems, Hoffman Estates, IL) equipped with a low-energy high-resolution parallel-hole collimator. To avoid artifacts caused by movement, mice were anaesthetized with 0.09 mg pentobarbital (Nembutal; Sanofi, Munich, Germany). Images were analyzed visually.
Uptake of FDG In Vivo
Thirty minutes after scintigraphy with 99mTc-MIBI, approximately 20 MBq glucose analog FDG (in-house production) were administered by intraperitoneal injection to 12 SCID mice; three carried an HT-29par tumor and nine carried an HT-29mdr1 tumor. To monitor the time course of FDG uptake by the tumor, a dynamic series of 12 frames (duration 5 min each) was acquired immediately after tracer application using a conventional, full-ring, whole-body PET (ECAT EXACT 921/47; CTI/Siemens, Knoxville, TN). Transversal images with pixels of 0.7 × 0.7 mm2 were reconstructed by filtered backprojection using a Shepp-Logan filter with a cutoff of 1.0 in units of the Nyquist frequency, leading to an in-plane resolution of 6.5 mm (full width at half maximum). Photon attenuation within the mice was neglected. To evaluate the uptake of FDG, the last four frames of the dynamic series were added, corresponding to a static acquisition from 40 to 60 min past injection. Transversal, coronal, and sagittal slices were analyzed visually on the computer monitor. To enable direct visual comparison with the results of the planar 99mTc-MIBI scintigraphy, maximum intensity projections from the posterior view were computed from the tomographic FDG PET slices.
Histology
After completion of the scintigraphy and PET, animals were killed by cervic dislocation. Tumors were excised within their capsule and cut into two pieces, one of them for paraffin wax histology and the other for electron microscopy. For paraffin histology, tissue specimens were immersion-fixed in 4% paraformaldehyde in 0.1 mol/L phosphate buffer (48 h, 4°C, pH 7.4), rinsed in 0.1 mol/L phosphate buffer (4°C, 12 h), dehydrated routinely, and embedded in paraplast using xylene as an intermedium. 5-μm coronal sections were mounted, rehydrated, and rinsed twice in PBS.
Immunohistochemistry.
HT-29par and HT-29mdr1 cells cultivated on Chamber Slides and fixed thereafter (see above), as well as rehydrated paraffin sections of tumors grown in SCID mice, were processed for immunohistochemic visualization of P-gp. After preincubation for 20 min in 10% normal swine serum diluted in antibody diluent (S-3022; Dako, Carpinteria, CA), slides were incubated for 1 h (20°C) in primary monoclonal antibody JSB-1 (No. MON 9011; Sanbio, Beutelsbach, Germany), diluted 1:100 in antibody diluent. After three rinses in 0.05 mol/L tris buffered saline ([TBS] pH 7.6), the tissue was incubated in biotinylated secondary antibody (LSABR2 Kit, K-0674; Dako, Carpinteria, CA) for 10 min (20°C), rinsed three times in TBS, and then incubated in streptavidin-alkaline phosphatase complex (LSABR2 Kit, Dako) for 30 min. Alkaline phosphatase activity was visualized after three washings in TBS by incubation in the substrate naphthol-AS bisphosphate supplemented with hexatozised New Fuchsine (Code No. 22931–8; Aldrich, Steinheim, Germany) for simultaneous coupling (20 min, room temperature). In addition, antilaminin (L-9393; Sigma, Steinheim, Germany) and antifactor VIII (Clone M 0616; Dako, Glostrup, Denmark) immunohistochemistry were used to visualize tumor blood vessels after the same staining protocol. The specificity of the immunoreactions was assessed by omission of either the primary or secondary antibody or the streptavidin-alkaline phosphatase complex, each resulting in negative staining. Slides were then rinsed, counterstained with hematoxylin, and mounted with Crystal Mount/Clarion (Biomeda).
Electron Microscopy.
For ultrastructural examination, tissue specimens of about 4 mm3 were taken from the tumors grown in SCID mice and immersed in a fixative consisting of 6% glutaraldehyde in 0.05 mol/L phosphate buffer (24 h, pH 7.3, 760 mOsm). The tissue was rinsed in a mixture of 0.1 mol/L phosphate buffer and 0.1 mol/L saccharose, postfixed with buffered osmium tetroxide (1% for 2 h), dehydrated in ethanol, cleared in propylene oxide, and embedded in glycidether 812 (Serva, Heidelberg, Germany). Then 1-μm-thick transverse semithin sections were cut with an OMU 3 ultramicrotome (Reichert, Vienna, Austria) and stained with toluidine blue and pyronine red. Ultrathin sections (70–90 nm) were prepared for electron microscopy and stained with saturated alcoholic uranyl acetate and lead citrate before examination with a CM100 transmission electron microscope (Phillips, Hamburg, Germany).
RESULTS
In Vitro Studies
Immunohistochemistry.
P-gp expression of cell lines grown on coverslips was visualized by the use of the monoclonal antibody JSB-1. All HT-29par cells were completely devoid of the anti-P-gp label (Fig. 1A). In contrast, almost all HT-29mdr1 cells were labeled distinctly, indicating the presence of P-gp (Fig. 1B). However, the intensity of the JSB-1 immunoreaction varied between individual cells. The majority of HT-29mdr1 showed a moderate to strong staining, few cells were weakly positive, and <5% of the cells were completely unlabeled. The JSB-1 immunoreaction was most marked on the surfaces of the HT-29mdr1 cells.
In vitro characteristics of HT-29par (A and C) and HT-29mdr1 (B and D) cell lines. Immunohistochemic visualization of P-gp expression: absence of label in HT-29par (A) and positive staining in HT-29mdr1 (B) cell lines. Functional assessment of P-gp was performed by Rhodamine 123 accumulation (C and D). Note bright fluorescence in HT-29par (C) cells in contrast to weak fluorescence in HT-29mdr1 cells (D) caused by increased efflux. Original magnification, ×450.
Rhodamine 123 Uptake.
MDR transporter activity was shown functionally by applying the fluorescent dye Rhodamine 123, which is taken up and retained by nonresistant cells but pumped rapidly out of the cytoplasm by the P-gp transporter protein in MDR cells (15). All HT-29par cells exhibited a bright fluorescent staining of varying intensity (Fig. 1C). In contrast, HT-29mdr1 cells showed only a very weak fluorescence; very few cells (≤5%) exhibited strong Rhodamine 123 staining (Fig. 1D). Additionally, inhibition of the MDR transporter function by adding Tween 80 to the Rhodamine 123 incubation medium resulted in a marked increase in HT-29mdr1 cell fluorescence.
99mTc-MIBI Uptake.
The MDR transporter function was assessed quantitatively in vitro by measuring the uptake of 99mTc-MIBI, which was significantly different in HT-29par and HT-29mdr1 cells (P < 0.0001). In HT-29par cells, 0.40% ± 0.09% (mean ± SD) of the activity of the external medium was detected after an incubation period of 1 h (Fig. 2). In contrast, in HT-29mdr1 cells, only 0.05% ± 0.01% was observed. 99mTc-MIBI uptake was therefore approximately eight times higher in HT-29par cells than in HT-29mdr1 cells.
Uptake of 99mTc-MIBI and 201TlCl in HT-29par (black columns) and HT-29mdr1 (white columns) cells in percentage tracer in external medium. Data represent mean ± SD of nine cell suspensions each.
In contrast, the uptake of 201TlCl, which is not a substrate of P-gp, was not significantly different in both cell lines (P = 0.48); 1.38% ± 0.12% and 1.34% ± 0.13% of the activity of the external medium were detected in HT-29par and HT-29mdr1 cells, respectively (Fig. 2). Both 99mTc-MIBI and 201TlCl uptake essentially remained unchanged after incubation periods ranging from 1 to 4 h.
In Vivo Studies
Planar scintigraphic images of the SCID mice showed the physiologic distribution of 99mTc-MIBI, that is, uptake in the liver, urinary bladder, and intestine (Fig. 3). In addition, the injection site at the tail was visible in several animals. No significant difference in tracer uptake was observed between mice carrying HT-29par tumors (Fig. 3A) and those carrying HT-29mdr1 tumors (Fig. 3B). In addition, images obtained 15 and 60 min after injection essentially were the same (Figs. 3B and C). Neither HT-29par nor HT-29mdr1 tumors could be detected in any of the animals at any time examined.
Planar scintigraphic images from posterior view obtained 15 min after injection of 99mTc-MIBI in SCID mice carrying HT-29par (A) or HT-29mdr1 (B) xenografts between scapulae. Imaging of SCID mice carrying HT-29mdr1 tumors was repeated 60 min after injection (C). Upper threshold of color table was set to five times average tracer uptake in background. Physiologic tracer uptake is shown in liver, urinary bladder, and intestine. Tumors are not visible.
In contrast, tumors were clearly visible in FDG PET (Fig. 4). All HT-29par tumors and six of nine HT-29mdr1 tumors showed increased uptake of the glucose analog 60 min after injection, excluding the lack of tumor cell viability as a reason for negative 99mTc-MIBI imaging. Two of three HT-29par tumors showed a very high uptake of FDG, whereas the uptake in the remaining HT-29par tumor and in the HT-29mdr1 tumors for which the PET findings were positive was only moderately increased.
Maximum intensity projections from posterior view computed from transverse tomographic PET images acquired 50 min after injection of FDG in SCID mice carrying HT-29par (A) or HT-29mdr1 (B) xenografts between scapulae. Uptake of glucose analog was increased significantly in all HT-29par tumors and in six of nine HT-29mdr1 tumors (arrows).
Histology
Both HT-29par and HT-29mdr1 tumors examined after the imaging procedures were moderately differentiated, contained numerous mitoses, and displayed a sizable central necrosis. As visualized by antilaminin and antifactor VIII immunohistochemistry, both HT-29par and HT-29mdr1 carcinomas were generally poorly vascularized. A marked regional variation in blood vessel distribution was observed, however (Fig. 5); blood vessels were located mainly in the tumor periphery close to the capsule, whereas very few vessels were found in its center. Immunohistochemic visualization of the P-gp showed an almost complete absence of JSB-1 immunoreaction in all HT-29par tumors examined; <1% of the cells were weakly stained (Fig. 6A). In HT-29mdr1 carcinomas, a considerable variation between different cells in the JSB-1 label was observed. Overall staining intensity was much stronger than that of HT-29par tumors, however (Fig. 6B). In addition, as already observed in the cell culture, the intensity of the immunoreaction varied markedly between individual HT-29mdr1 cells in solid HT-29mdr1 tumors grown in SCID mice. Most HT-29mdr1 tumors were composed of cells showing weak, moderate, or strong JSB-1 immunoreaction; <5% of the cells were unlabeled. Few HT-29mdr1 carcinomas, however, contained up to 30% unlabeled cells. Generally, staining was observed in the cytoplasm of HT-29mdr1 cells. In addition, the membrane of numerous tumor cells was more strongly labeled than the cytoplasm. In differentiated areas of HT-29mdr1 and also in HT-29par tumors, tumor cells were forming tubules with visible lumina. Remarkably, JSB-1 staining was confined to the luminal surface of these tubules.
Visualization of HT-29par tumor vasculature basement membranes by antilaminin immunohistochemistry. Tumors grown in SCID mice show numerous blood vessels in their periphery (arrowheads) but very few in their center (arrow). Original magnification, ×450.
Immunohistochemic demonstration of P-gp expression in xenografts grown in SCID mice. Note absence of label in HT-29par (A) and positive staining in HT-29mdr1 (B) tumors. Original magnification, ×450.
Electron microscopy revealed that HT-29par tumor cells showing almost no 99mTc-MIBI uptake in vivo contained only a few mitochondria (Fig. 7A), whereas hepatocytes of the SCID mouse, displaying high 99mTc-MIBI uptake in vivo, were packed densely with mitochondria (Fig. 7B).
Electron micrographs of HT-29par xenografted tumor cells (A) and SCID mouse hepatocyte (B). Note paucity of mitochondria (asterisks) in HT-29par tumor cells and their abundance in SCID mouse hepatocytes. Original magnification, ×18,000.
DISCUSSION
This study compared in vitro and in vivo characteristics of the MDR colon carcinoma cell line HT-29mdr1 with those of the nonresistant tumor cell line HT-29par. Before implantation into SCID mice, the MDR of the HT-29mdr1 cell line was ascertained in vitro by three different techniques: immunohistochemic visualization of the P-gp epitope using the JSB-1 antibody (18), functional assessment of the MDR efflux pump by Rhodamine-123 uptake studies (15,19), and quantitative evaluation of the P-gp drug transporter activity by measuring 99mTc-MIBI accumulation (4).
Our immunohistochemic data show that HT-29mdr1 and HT-29par cell lines can be clearly distinguished immunohistochemically by binding of the JSB-1 antibody directed against the P-gp. Virtually all HT-29mdr1 cells are immunoreactive, whereas a complete absence of label is observed in HT-29par cell lines. The specificity and sensitivity of the monoclonal antibody JSB-1 for detecting the P-gp on paraffin-embedded sections have been shown by Pavelic et al. (18). JSB-1 immunostaining was applied successfully in numerous studies to detect MDR cells (19–21) and, in some tumors, to predict the response to chemotherapy (22). Consistently, positive JSB-1 antibody plasma membrane staining for HT-29mdr1 and the lack of immunoreactvity for HT-29par cell lines were reported previously by Spoelstra et al. (23). We found no differences between individual nonresistant HT-29par cells, which are all devoid of immunoreaction. However, the intensity of JSB-1 immunostaining varies between different HT-29mdr1 cells, indicating that all MDR cells express the P-gp, but to a variable degree.
This interindividual variation of P-gp expression in HT-29mdr1 cells was also shown in our Rhodamine 123 accumulation studies. The fluorescent compound Rhodamine 123 is a known substrate for P-gp (19), and this property is exploited to detect MDR function using fluorescence microscopy. Rhodamine 123 staining in drug-sensitive cells is manifold higher than in MDR cells (15,19); it is not determined by the initial dye uptake, but rather by an efflux process (24). This active extrusion is directly correlated with the expression of P-gp, as observed in 58 cell lines in the National Cancer Institute drug screen (25). Thus, Rhodamine 123 accumulates passively within cells, driven in part by the negative plasma membrane potential (17,26), and drug-sensitive cells retain the dye, thereby remaining fluorescent for many hours. In drug-resistant cells, however, Rhodamine 123 is extruded by the P-gp, resulting in negative staining (15,24). The use of cytoplasmic exclusion of Rhodamine 123 as a functional assay for detecting the activity of the P-gp efflux pump in vitro has become a generally acknowledged technique in MDR research (27,28). In addition, recent clinical studies indicate that the Rhodamine 123 efflux assay is also of prognostic significance to predict the response to chemotherapy in acute leukemia (28,29). We have shown that HT-29mdr1 tumor cells show essentially no accumulated fluorescence when compared with parental nonresistant cells, which are brightly fluorescent. These findings are in good agreement with previous studies showing that Rhodamine 123 efflux is markedly higher in HT-29mdr1 than in HT-29par colon carcinoma cells (11,12).
To exclude other conceivable mechanisms of reduced dye accumulation, that is, reduced permeability of the cell membrane and reduced binding of the dye to the intracellular targets, we inhibited the MDR efflux pump by Tween 80 (15), which leads to a cessation of Rhodamine 123 efflux and accumulation of the dye inside the cell. HT-29mdr1 cells incubated with Rhodamine 123 in the presence of Tween 80 are brightly fluorescent, confirming that the decreased fluorescence of HT-29mdr1 cells is indeed mainly caused by the presence of the MDR efflux pump.
Another, even more clinically relevant, substrate of the P-gp transporter pump, 99mTc-MIBI, was used in our study as a quantitative measure of MDR function. Unidirectional uptake of 99mTc-MIBI is driven thermodynamically by negative plasma membrane potentials and mitochondrial inner matrix potentials, thereby concentrating the agent within the cells in a manner similar to that of other lipophilic cationic probes of membrane potential (4). 99mTc-MIBI uptake is linearly related to cell number and is proportional to the extracellular concentration of 99mTc-MIBI over a range of 4–2800 pmol/L (17). The total concentration of Tc-MIBI (i.e., a mixture of 99Tc-MIBI and 99mTc-MIBI as obtained from the 99Mo/99mTc generator) administered in our study (20 pmol/106cells) is within this range. Kinetic studies show that 99mTc-MIBI uptake approaches a plateau at 30 min and reaches a maximum level of uptake at 1 h, which it maintained for at least 3 h thereafter (17). Similar results have been obtained in our studies. In vitro 99mTc-MIBI uptake varies between different tumor cell lines. A range from 5% of the activity in the external medium in the differentiated human hepatocellular carcinoma to 28% in the human breast carcinoma cell line BT-20 has been reported (17). 99mTc-MIBI uptake in our nonresistant HT-29par colon carcinoma cells is 0.4% of the activity of the external medium, which is a very low percentage when compared with other tumor cell lines.
Being a substrate of the P-gp efflux pump, 99mTc-MIBI is actively transported out from MDR cell lines (4). Enhanced extrusion results in reduced 99mTc-MIBI accumulation in chemoresistant cells by 10- to 200-fold, thereby allowing detection of the MDR phenotype. Our observations that 99mTc-MIBI uptake by HT-29mdr1 cells is about 8 times lower than that of HT-29par cells is in accordance with these results. Because 99mTc-MIBI is an established tracer for scintigraphy, it has the potential of imaging P-gp in vivo, thereby predicting the outcome of chemotherapy (6,7,9,21,30). In fact, 99mTc-MIBI has been used successfully to visualize P-gp–mediated efflux in an animal tumor model (4). One major drawback in this study was the finding that even nonresistant HT-29par tumors grown in SCID mice could not be detected by 99mTc-MIBI scintigraphy. Tumors were not visible 15 or 60 min after 99mTc-MIBI injection, which is within the usual time frame reported for 99mTc-MIBI scintigraphy (4,6,21). Several mechanisms underlying this unexpected observation are conceivable. Lack of viable tumor cells cannot account for the negative imaging results, because HT-29par tumors were clearly visible in FDG PET. Another reason to consider is poor tumor vasculature, because 99mTc-MIBI uptake is dependent on sufficient blood supply (20). The relatively low density in blood vessels observed histologically in HT-29par tumors may therefore contribute to the negative imaging. However, the low percentage 99mTc-MIBI uptake by HT-29par tumor cells observed in vitro before implantation into SCID mice indicates that tracer accumulation by the nonresistant tumor cells themselves is not high enough to allow detection by scintigraphy. Successful scintigraphic imaging has been reported mainly in breast cancer (9,31). Cell lines derived from this tumor show a high 99mTc-MIBI uptake in vitro (17). Clinical data based on a heterogeneous group of tumors are, however, difficult to compare with defined tumor cell lines, because 99mTc-MIBI uptake is influenced markedly by the histologic type of the cancer (31). The only study comparing 99mTc-MIBI uptake of cultured cells with that of tumors grown in nude mice (4) examines different cell lines in vitro and in vivo. The poor 99mTc-MIBI uptake by HT-29par tumor cells may be attributable to P-gp expression by this nonresistant cell line. This explanation is unlikely, however, because we did not observe positive JSB-1 immunoreaction in vitro or in the solid HT-29par tumors grown in SCID mice. However, different cellular resistance mechanisms other than P-gp expression, which have been discussed for colorectal cancer (32), cannot be excluded, for example, MDR-associated protein expression (5,26).
Because 99mTc-MIBI uptake is dependent mainly on the electrostatic gradient of mitochondrial membranes and 99mTc-MIBI therefore accumulates in tissues rich in mitochondria (16), we believe that the scarcity of mitochondria is one of the principal reasons for poor tracer uptake. Very few mitochondria have been observed in electron micrographs of cultured HT-29par tumor cells (33). This study shows that the mitochondrial density of HT-29par cells in solid tumors grown in vivo is much lower than that of hepatocytes of the SCID liver, which shows high 99mTc-MIBI uptake in scintigraphy. This assumption is also corroborated by our measurements of 201TlCl uptake. Being 1.4% of the activity in the external medium, the accumulation of 201TlCl, which also distributes across mitochondrial membranes (34) but is not a substrate of the P-gp transporter, has also been relatively low compared with 5% reported for other tumor cell lines (35). Our results show that sufficient 99mTc-MIBI uptake by nonresistant cells, which is related to their content of mitochondria, is the major prerequisite for successfully distinguishing between MDR and drug-sensitive cells. Thus, negative 99mTc-MIBI scintigrams are not necessarily associated with P-gp expression.
One unexpected finding of this study was the reduced FDG uptake of HT-29mdr1 tumor cells compared with HT-29par cells observed in PET studies. Up to now, PET studies designed to diagnose MDR in vivo have used radiolabeled substrates of the P-gp transporter, for example, colchicine (36), verapamil (30), and daunorubicin (30). The only study examining MDR tumors using FDG (36), although describing a decreased 11C-colchicine uptake in drug-resistant human neuroblastoma xenografts, does not report any differences in FDG accumulation. The decreased FDG accumulation observed in our study is not attributable to differences in tumor volume, because care was taken to let tumors grow to approximately the same size. It also cannot be explained by a larger central necrosis, which could be excluded by our histologic analysis. It is also unlikely that diminished FDG uptake of HT-29mdr1 tumor cells is caused by decreased energy metabolism. The P-gp transporter is a drug efflux pump dependent on adenosine triphosphate, bringing about greater energy demand by MDR cells (37). In fact, MDR is associated with an elevated rate of glycolysis (38) and a higher glucose requirement (39). Therefore, it seems most likely that FDG accumulation is reduced because of an altered glucose transport into HT-29mdr1 tumor cells. Plasma membrane glucose transporter GLUT-1 levels are known to be diminished progressively with elevated P-gp levels (37). This result may also lead to reduced FDG uptake visible in PET imaging. Moreover, differences in glucose metabolism have been described in MDR cell lines (40), which may contribute to decreased FDG accumulation. Our results indicate that FDG PET may therefore be a potential marker for detecting P-gp in vivo. A systematic analysis of FDG accumulation by HT-29par and HT-29mdr1 tumor cells in vitro and in vivo is currently in process.
CONCLUSION
In this study, P-gp expression of nonresistant and MDR HT-29 human colon carcinoma cell lines was determined systematically both in vitro by immunohistochemic and functional assays and in vivo by functional imaging using 99mTc-MIBI scintigraphy and FDG PET of the same carcinoma cells growing in SCID mice. Results were compared with histologic and electron microscopic analyses of the xenografted tumors. Our results show that sufficient 99mTc-MIBI uptake by nonresistant cells, which is related to their content of mitochondria, is the major prerequisite for successfully distinguishing between MDR and drug-sensitive cells. Thus, negative 99mTc-MIBI scintigrams are not necessarily associated with P-gp expression. In some tumors, FDG PET may be an in vivo marker for MDR.
Acknowledgments
The authors thank Klaus Desler, Susanne Feldhaus, Sibylle Leich, and Klaus Siebert for excellent technical assistance and Dr. Jochen Düllmann for critically reading the manuscript.
Footnotes
Received Aug. 30, 2000; revision accepted Nov. 17, 2000.
For correspondence or reprints contact: Karl H. Bohuslavizki, MD, PhD, Department of Nuclear Medicine, University Hospital Eppendorf, Martinistrasse 52, D-20246 Hamburg, Germany.